The present invention relates to noise in digital communication systems, and more particularly, to distinguishing particular classifications of noise in a digital communication system and using those distinctions to determine an accurate noise margin.
The reliability of a digital communication channel is commonly expressed in term of Bit Error Rate (referred to herein as “BER”). A digital communication system is typically designed to provide a reliability level better than some worst-case reference level, further referred to as BERreq, dependant on the type of service provided above this communication channel.
Digital signal processing theory shows that the bit error rate of a communication system is a function of the Signal power to Noise power Ratio (referred to herein as “SNR”) at the input of the receiver. Let us call SNRreq the required SNR necessary to achieve a BER equal to BERreq. In order to provide a good quality of service, it is common practice to require that a communication system operate at an SNR exceeding SNRreq by some factor known in the art as Noise Margin (referred to herein as “m”). The Noise Margin is defined as the amount of external noise increase that the communication system can tolerate while still insuring a data transport with a BER lower than BERreq. The Noise Margin m may be calculated as:
  m  =      SNR          SNR      req      The noise margin is typically expressed in a logarithmic scale (e.g., in dB).
The noise present in a communication system can be classified according to its source. For example, the noise can theoretically be distinguished as:                The external noise (Ne) defined as the noise already present on the signal at the input of the receiver.        The internal noise (Ni) defined as the equivalent noise increase introduced by the non-ideal behavior of the receiver. This typically includes the receiver input noise, analog to digital converter noise and non-linear behavior, residual echo noise in duplex systems, residual inter-symbol interference, etc. . . .        
The noise margin required from a communication system only applies to the noise sources that are subject to variation over time. In many systems, the internal noise can be safely considered as being constant over the time, and therefore does not figure into the noise margin calculation. However, for the sake of simplicity, in many communication systems no distinction is made between internal and external noise in the computation of the noise margin. One disadvantage to computing noise margin this way is that the resulting noise margin may be unnecessarily large, which translates to sub-optimal system performance in areas such as data throughput, range or power consumption.
It is an object of the present invention to substantially overcome the above-identified disadvantages and drawbacks of the prior art.